First let me warn you I have no academic formation in Linguistics, I can't define that area well, so if this sounds off-topic, it probably is.

"I know what you know" is an ambiguous sentence, carrying two distinct meanings: that "I" am in knowledge of, at least, the very same things "you" are; or that "I" have information of which subjects "you" know (without this meaning "I" know anything about subjects "you" knows).

I have several related questions associated to the sentence: is there any way to separate the two meanings without a full explanation (kind of what I did)? What about other languages? In Portuguese, the same thing happens. Esperanto often seems to have a way to carry just the needed precision, yet I can't think of any way to distinguish those meanings right now. With what types of verbs does this work? For instance, it doesn't work with "think" (a proposition is required in at least in one of the cases) and most certainly not with "eat".

  • 2
    I'm not 100% sure, and i'm a little confused by your description of the two different readings, but i think what you've identified is the de dicto vs. de re readings of an embedded interrogative. The first reading that you identify would correspond to the de dicto (lit. "the word") reading - 'what you know' isn't referring to a specific proposition, but just whatever the referent happens to know. The second reading is de re (lit. "the thing") - 'what you know' is referring to a specific proposition. Can anyone confirm/disconfirm this?
    – P Elliott
    Aug 9, 2013 at 23:39
  • 1
    Yes, they're somewhat distinguishable. There are two varieties of such embedded question clauses, which Ross calls Conjunctive and Disjunctive. A conjunctive clause what you know refers to 'A and B and C and D and ...', while a disjunctive clause what you know refers to 'A or B or C or D or ...'. The subject complement in What she knows comes as a big surprise is conjunctive, but the subject complement in What she knows is still a big mystery is disjunctive.
    – jlawler
    Aug 10, 2013 at 0:07
  • 1
    Here's one way of making the difference a little clearer. Consider the following premises: (a) John knows what Bill knows. (b) Bill knows nothing. Under one reading, the conclusion (c) is licensed: (c) John knows nothing. Under the second reading, the conclusion in (d) is licensed: (d) John knows that Bill knows nothing.
    – P Elliott
    Aug 10, 2013 at 0:20
  • 1
    Having thought about this some more, i think i was wrong to identify the ambiguity as a matter of de re vs. de dicto. Rather, i think that what we have hear is a structural ambiguity between 'know' with a free -relative as its complement (syntactically a DP), and 'know' with an indirect question as its complement (syntactically a CP). The free relative reading corresponds to the conclusion (c), and the indirect question reading corresponds to the conclusion (d). I'll try to spell this out some more in an answer if i have a spare 20 mins at some point.
    – P Elliott
    Aug 10, 2013 at 12:22
  • 1
    Yes. That's another way of stating the distinction that Ross makes. Relative clauses are presupposed material, hence conjunctive; questions are unknown material, hence disjunctive. But there's no syntactic difference between a free relative (or headless relative, or fused relative) clause and an embedded question clause. When you leave out that much stuff in a clause, you lose the ability to mark that distinction, is essentially what I think happens. The only thing left is the semantic distinction, and that's usually interpreted by context.
    – jlawler
    Aug 10, 2013 at 14:01

2 Answers 2


I would submit that prosody could play a role in disambiguating the two readings, at least in English. As a native speaker, if I heard someone say (out of the blue):

1) I know what you know.

I would interpret it as, "I have knowledge about the same things you have knowledge about." On the other hand, if I heard:

2a) I know what you know.


2b) I know what you know.

I would interpret it as, "I know which subjects you have knowledge about."

The reason is that we tend to emphasize things that we are comparing or contrasting entities in parallel constructions (a phenomenon known as contrastive focus) and to de-emphasize things that are repeated or taken for granted in the context of the conversation (things that are given or presupposed).

In (1), the two instances of know are referring to the same knowledge, so they are de-emphasized, but the pronouns (i.e. the "knowers") are different, so contrastive focus is utilized. This version of the utterance would have the same prosodic pattern as something like, "I watched what you watched" (in response to the question, "What did you watch on TV last night?").

Emphasis on the last content word in a phrase or a clause (without getting theory-specific here) can also signal focus on that whole constituent (known as broad focus). In both (2a) and (2b), the constituent "what you know" is being focused as new information, and so the last word, know, gets emphasized. Whether the first know is emphasized or not could depend on the speech rate as well as the context. In faster speech or in a context where the topic of the speaker's knowledge is given (e.g. if the speaker is answering the question "What do you know?") the first know may not get emphasized. But in more careful speech or in a context where the topic of the speaker's knowledge hasn't been established (e.g. if the speaker is responding to the question "Why are you smiling?") the first know may get emphasized. The prosodic patterns given in (2a) and (2b) would be appropriate for something like "I ate what you cooked" (in response to "What did you eat for dinner tonight?" or "Why did you get sick last night?").


I agree with P Elliot above, that your two sentences are:

with regards fact "X" (as an example, "swans are white")

I know X, that you know as well. (I know that swans are white too.)
I know that you know X (but may not know it myself). (I know that you know what colour swans are.)

(As has been pointed out in comments, purely syntactically, the first version of "I know what you know" implies that I share all of your knowledge, not just a specific fact. However, I feel that in use, nobody would ever suggest that they know everything another person knows, and so there must be an implied context to the use of this sentence. The discussion around the differences between the two sentences, and other potential cases of this ambiguity still apply whether we are talking about a specific context or a general totality of knowledge, so this point is largely irrelevant to the rest of the discussion.)

In thinking about these two a little more, I came up with this, that I think also explains the ambiguity well:
1) I can answer [a question/all the questions] that you can.
2) I know [a question/all the questions] that you can answer.

In speaking each of the sentences, it would be emphasis that would separate the two. "I KNOW what you know" is different from "I know WHAT you know."

Besides re-phrasing, which could be done in many ways, an example above in this answer, I can't think of how you could distinguish the two in writing.

I am trying to think of how to classify verbs where this kind of ambiguity, and coming up blank. The reason that it works with "know" is that the object of "know" is a fact, and "the thing you know" and "that you know a thing" are both facts. Of course, with any verb, if the object of the verb, call it "X" is a "Y" then "the thing that you X" must also be a "Y", but "that you X a thing" is always going to be a fact. So a verb that can be applied to a fact in this manner, should be a possible candidate for this ambiguity.

As you pointed out, it doesn't work with "think", because although you can "think" a fact, the phraseology is somewhat different. "I think what you think" (I think like you) vs "I think that you think [that]." (I know what you think) Attempting to force the sentences into the same structure to allow the ambiguity brings us back to using the word "know".

One other word I've found that works is "say". "I say what you say" can mean "I say the same thing as you" or "I tell you what to say".

"I suspect what you know" also has similar ambiguity - but only because suspect basically means "think I know" so the sentence "I think I know what you know" still contains your original sentence.

I can't help with other languages, but I suspect that any with common forms in the declension of a noun, or a specific form for a verb when it's used in a noun phrase, will have this same feature.

  • I have to disagree with your agreement. "I know what you know" implies I do not only know something you do, but all of it. It's like saying "those chairs are brown" implies all of those are brown, not some. As such, "I know what you know" is never "I know a thing that you know" but "I know the thing**s** you know", and, unfortunatly, the plural makes it as unclear as the first sentence. The same happens with "I know that you know a thing", except this one is invalid at least twice. I may know you know a thing, but not what thing.
    – JMCF125
    Aug 10, 2013 at 15:44
  • If all chairs are brown is true, and there exists a chair, then there exists some chair that is brown. This is basic.
    – jlawler
    Aug 10, 2013 at 16:38
  • 1
    My own native intuition is that it's still ungrammatical (or at least infelicitous) even with "or not" appended to the end. We can create some nice minimal pairs here, since think about does seem to sub-categorise for Q-complement: (i) *Are you thinking whether John knows something? (ii) Are you thinking about whether John knows something? (iii) Are you thinking what i'm thinking (unambiguous - conjunctive interpretation only). (iv) Are you thinking about what i'm thinking (ambiguous - conj. and disj. interpretations both available).
    – P Elliott
    Aug 11, 2013 at 17:53
  • 1
    (i-iv) above are meant to show that the relevant ambiguity only arises when the predicate sub-categorises for a Q-complement. (i) shows that think doesn't take a Q-comp. (ii) shows that think about does. (iii) shows that the ambiguity doesn't arise with think, (iv) shows that think about is unambiguous - patterning with the availability of a Q-complement.
    – P Elliott
    Aug 11, 2013 at 17:55
  • 1
    "(iv) think about is unambiguous" <- think you meant IS ambiguous. The ambiguity is clearer to me with another "about" on the end of (iv) but I get what you mean. "Are you thinking" seems to behave very differently to "Do you think", interestingly. "Do you think whether John knows something" doesn't scan at all "Do you think about whether John knows something" does. Both versions seem to work with "Are you thinking"
    – Ryno
    Aug 12, 2013 at 1:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.